Imatinib is the first-line treatment for gastrointestinal stromal tumors; however, the clinical prognosis and adverse reactions of patients vary owing to individualized discrepancies in plasma exposure. To determine the safe interval for steady-state plasma trough concentrations (C min ) of imatinib and its active metabolite, N-demethyl imatinib (NDI), 328 plasma samples from 273 patients treated with imatinib were retrospectively analyzed. Imatinib C min and NDI C min were tested, and adverse reactions were recorded. The association between imatinib C min , NDI C min , and serious adverse reactions was evaluated. The C min range of imatinib was 209.5-4950.0 ng/mL, with the mean value and SD of 1491.8 ± 731.4 ng/mL. The C min range of NDI was 80.0-2390.0 ng/mL with the mean value and SD of 610.8 ± 281.5 ng/mL. NDI C min was positively correlated with imatinib C min , whereas the ratio of NDI C min to imatinib C min (NDI C min /imatinib C min ) was negatively correlated with imatinib C min . Univariate logistic regression analysis demonstrated that the treatment objective, daily dose, imatinib C min , NDI C min , and imatinib C min + NDI C min were significantly associated with serious adverse reactions. Multivariate logistic regression analysis showed that NDI C min was an independent risk factor for serious adverse reactions, with a threshold of 665 ng/mL. NDI C min was an independent risk factor for serious adverse reactions, with a threshold of 665 ng/mL. Monitoring NDI C min was beneficial for the rational application of imatinib and individualized treatment of patients with gastrointestinal stromal tumors.